More screenshots
Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+21 moreClaude CodeCursorCodex CLIMuseDots
Manus
Grok
Grok Bot+18 agentsSearch Recovery Workbench
Turn normalized Search Console exports and page evidence into a traceable recovery ledger, scored backlog, reversible implementation drafts, and a measurement plan - offline and human-reviewed.
$15
Search Recovery Workbench
Example session with this skill installed
Diagnose our CTR drop. Here is a normalized two-period Search Console CSV (period, page, query, clicks, impressions, position) and the bookkeeping service page copy. Give me evidence-linked tickets I can hand to a developer.
- Read your context and instructions
- Compiled the search recovery workbench
SYNTHETIC DEMO - all data below is synthetic and labelled synthetic-demo. It is not a client result or a business outcome.
Scope: CTR Recovery + Conversion Gap + Implementation Plan on example.test, two equal 28-day periods, 10 CSV rows, 5 URLs known, 1 page fixture actually read, 0 live pages inspected.
Evidence-backed finding (E0005-E0008): the bookkeeping service page fell from 100 clicks / 1,200 impressions (CTR 8.33%) to 50 clicks / 1,200 impressions (CTR 4.17%), a -4.17 percentage-point CTR delta. This is a signal to investigate, not proof of a cause - demand shifts, SERP features and query mix remain unknown.
Backlog ticket C001 - Clarify the promise and next step (Medium, score 74.0; factors I4 E4 C3 F2 D1 R1): the fixture hero says "Grow your business" and its CTA points to an unrelated deck, while the body offers bookkeeping support for small retailers. Action: review the separate dry-run draft title, description, H1 and CTA; confirm the real contact process before applying anything. Success metric: qualified enquiries per eligible session (baseline not available). Status: recommended, not applied, outcome not measured.
Ticket C002 - Collect more evidence before rewriting the payroll page (score 57.0): only 20 and 18 impressions, so the sample is too small for a CTR diagnosis. Category: data-needed.
Ticket C003 - Hold consolidation (score 55.0, blocked-review): one query appears on two URLs, which is an overlap candidate only. Do not merge or redirect before a SERP, intent and canonical review.
Every ticket ships with evidence IDs, acceptance tests and rollback notes. Nothing is edited or published automatically.
Connects securely to your tools. The creator never sees your data.
About this skill
Search recovery work is often scattered across a spreadsheet, a content brief, and a developer chat. Search Recovery Workbench creates a durable handoff: an evidence ledger, a scored backlog, dry-run patch drafts, acceptance tests, rollback instructions, and a dated measurement plan.
Use it with normalized CSV rows containing period, page, query, clicks, impressions, and position. The offline Python 3 standard-library engine validates data, calculates CTR and deltas, highlights small samples, and applies a transparent 0-100 triage formula. An agent handles approved manual page or sitemap review and records exactly what was inspected.
Nine guided modes cover Quick Diagnosis, bounded Full Website Opportunity Audit, Keyword-to-Page Mapping, Content Refresh, CTR Recovery, Conversion Gap Analysis, GEO Readiness Review, Implementation Plan, and Post-Change Measurement. Every output is a recommendation or draft. Human approval is required before any change.
What you receive
- Nine guided recovery modes
- Offline normalized-CSV analysis with repeatable CTR/delta calculations
- Evidence IDs, source locators, statuses, owners, and acceptance tests
- Explicit impact/evidence/confidence/effort/dependency/risk scoring
blocked-reviewfor high risk or dependency ≥4; zero evidence caps score at 39.9- Dry-run implementation drafts and per-change rollback documentation
- Baseline/post-change measurement with
not measuredwhen data is unavailable
Honest limitations No API integrations, live rank or keyword-volume data, auto-crawl, mass crawl, CMS editor, autopublish, or auto-apply. GEO review measures readiness only, not actual AI citations. Outputs are drafts and recommendations, not proof of recovery, revenue, ranking, or conversion improvement. Cross-host native compatibility has not been tested; the self-contained manual core in SKILL.md works without supporting files. Local QA (31 unit tests) passed; that is not a marketplace security approval.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 4 days ago
Requires